Convergence acceleration of logarithmic fixed point sequences
β Scribed by Paul Sablonniere
- Publisher
- Elsevier Science
- Year
- 1987
- Tongue
- English
- Weight
- 255 KB
- Volume
- 19
- Category
- Article
- ISSN
- 0377-0427
No coin nor oath required. For personal study only.
β¦ Synopsis
Let (x,)
be some sequence generated by x,+ 1 = f(x,) where
i>1
For x 0 > 0 small, it converges to zero logarithmically, i.e. lim, xn+l/x . = 1, thus we need algorithms for accelerating its convergence. Using asymptotic expansions in the analysis of the A 2 and 02-algorithms leads to modified iterated versions of the first one and to combinations with the iterated 02-algorithm. In particular some superconvergence phenomena can be explained in this framework. A similar study can be made for other nonlinear algorithms known at present. Moreover, the above algorithms are also good accelerators for large classes of slowly convergent integrals and series.
π SIMILAR VOLUMES
There are several convergence acceleration methods that are based on the evaluation of partial sums sn for relatively large n, and thus, normally require the evaluation of all terms aj with 0 < j < n. Here, we show that it is possible to avoid the computation of the partials sums of high order if it